This is an archive of the original site and you may encounter broken links and/or functionality

Xerte, Youtube and Graphs: Building an Education Graph in the Linked Data Age -- 129 -- Short (oral) Paper


10:30 - 11:30 on Thursday, 13 September 2012 in 4.204

Recent years have seen an explosion in the application of network- and graph-oriented perspectives to domains as diverse as social networks, fraud detection and drug discovery. Major Web properties such as Facebook and LinkedIn have built business models on capturing the social and professional graphs, respectively, while the Linked Data movement has spawned the emergence of a data commons on the Web based on a graph-oriented data model. In the field of education, learners, teachers, resources, courses, lecturers and institutions all form nodes in an interconnected network, but this graph is not systematically used at present to enhance the learning experience and outcomes of university students.

Our work aims to address this issue by assembling a sector-wide education graph from institutional sub-graphs that describe the relationships between courses and learning resources. These institutional sub-graphs take the form of learning resource lists created through Talis Aspire Campus Edition, a software-as-a-service application based on Linked Data principles and technologies, and currently in use at 30 universities in the UK and beyond. The Linked Data technology stack, combined with custom processes for citation parsing and coreference resolution, provides the ability to easily pool data from multiple institutions through a lightweight data warehousing-style process.

The result of this data linking and warehousing process is a unique view of the broader education graph, including hundreds of thousands of learning resources used on tens of thousands of university courses. As with traditional data warehousing, this Linked Data-based approach is enabling the development of sophisticated analytics over each institutional sub-graph. However, more notable insights are possible by examining cross-institutional patterns that give a unique insight into how learning resources are used across UK universities. For example, analysis of this education graph reveals the most widely used resources in each discipline. In addition, analysis of how items co-occur on resource lists enables the generation of recommended resources (and recommended lists) for independent learners or course developers.

By assembling an education graph based on courses and learning resources, this work has begun to reveal meaningful insights that are already contributing back to enhancing the learning and teaching process. In future work we plan to explore mechanisms for further enhancing, enriching and analysing this graph to support learners and teachers in UK institutions and beyond, and understand how a broadening and deepening of the education graph can uniquely enhance the educational experience.



Comments


Thanks David. You took forward our understanding of the potential of graphs.

Thursday, 13 September 2012, 12:07